What a relief last winter is finally over. In late December, California was hit by the first in a series of powerful storms called atmospheric rivers. These ribbons of extraordinarily wet air rush across the ocean and can dump staggering amounts of rain and snow upon landfall. After the driest three year stretch on record, it seemed like a miracle: Water! Falling from the sky!
Then eight more atmospheric rivers arrived in January. And they kept coming (and coming) through February and March — so many I lost count. Soon I wished the torrential rains would just stop. I felt like a bad Californian.
It is wonderful to see brimming reservoirs and towering snowpacks. But unrelenting storms also left floods, mudslides, evacuations, highway closures, and power outages across much of the state.
My elderly mom and her unhappy cat hunkered down in a hotel for days, waiting to see if her Monterey County neighborhood would flood. Friends with young children lost electricity and internet in their Santa Cruz Mountains home, and were trapped there by uprooted trees that fell across roads. A Bay Area friend was stranded in Santa Barbara and chronicled the many washed out roads, rock falls, and sinkholes that blocked her passage home in a series of ever more anxious Facebook posts.
While alternating between drought and deluge is nothing new for California, climate change is making these swings even more dramatic. New research will help the state prepare for future water extremes by tightening both long and short term forecasts, as well as optimizing water savings in the wettest years for use in the inevitable dry stretches. New policies could make water allocation more equitable during years like this one, when rivers are running high. And updates to the state’s water system will help safeguard deliveries to cities and farms as the supply boom and bust cycle grows ever wilder.
“Winter was weirdly wet,” says Michael Wehner, a climate and extreme weather scientist at Lawrence Berkeley National Laboratory and a lead author on the latest Intergovernmental Panel on Climate Change report. Scientists can’t say whether climate change caused the astonishing number of storms, he continues, but they can say it made them, “wetter and so that much worse.”
Researchers have long known that atmospheric rivers carry more water as temperatures rise, simply because warmer air holds more moisture. Now, Wehner and colleagues have discovered another reason downpours are intensifying. Climate change can also make storms release more of their contents, the team reported in a 2022 study of future precipitation in the San Francisco Bay Area. “Atmospheric rivers rain out more than expected — up to twice as much,” Wehner says.
This research was commissioned by the San Francisco Public Utilities Commission (SFPUC), which provides water, power and sewer services, and the findings will guide long-term infrastructure planning to protect the city from the stronger storms that are coming. “The study connected scientists on the hill with decision makers on the ground,” Wehner says, adding that this type of collaboration is a first for him. “It’s exciting.”
Not all atmospheric rivers rain more than anticipated. “The big storms are atmospheric rivers connected with extratropical cyclones, many of which are the bomb cyclone kind of storm,” Wehner says. A cyclone is an area of low pressure with wind sweeping around it. A bomb cyclone forms when the pressure falls precipitously, making the wind suddenly spin much faster.
Most atmospheric rivers in California coincide with extratropical cyclones. “They’re the most common type of storm event in California and are the most affected by climate change,” Wehner says. These combination storms are already at least 5% wetter than they would have been historically in the Bay Area, and his 2022 study showed that they could unleash up to 17% more rain by 2050 and up to 37% more by 2100.
Wehner wasn’t entirely surprised that atmospheric rivers are getting so much rainier because he’s seen a similar trend in hurricanes. A 2018 study he co-authored showed that in Katrina and other recent hurricanes, climate change increased rainfall as much as 10% compared to pre-industrial times. “The rain in the rainiest parts intensifies in both hurricanes and atmospheric rivers,” he says.
Wehner isn’t exactly sure why climate change makes atmospheric rivers drop more of their moisture, but he has some ideas. For example, these storms typically rain hardest when winds lift them uphill, where the pressure is lower. The strong winds of extratropical cyclones can provide that lift.
Winter extratropical cyclones will likely strengthen with climate change, with the windiest parts growing up to 40% bigger by the end of the century, according to a recent study.
Whatever the underlying cause of Wehner’s findings, the take homes for SFPUC are that rains in the San Francisco Bay Area will likely become even heavier than previously projected, and that rainfall in a changing climate is more complex than researchers had thought. “There’s so much we don’t know about how climate change affects extreme weather,” Wehner says.
While planners need to know what to expect over decades, emergency services providers need to know when and where storms will strike in real time. Anima Anandkumar, a computer scientist at the California Institute of Technology, is tapping artificial intelligence to speed weather forecasting.
Weather is caused by air moving in the atmosphere, which is in constant flux. Anandkumar was already modeling fluid dynamics — which is the flow of liquids and air — with artificial intelligence. “I thought if it can do that, why can’t it predict weather?,” she recalls. “So we just tried and it succeeded beyond our expectations.”
Today’s weather models are numerical: they take massive amounts of current weather data — temperature, humidity, wind and so forth — and plug them into mathematical equations that predict how air will move. Crunching all those numbers involves trillions upon trillions of calculations, requires supercomputers, and takes hours.
Traditional computer programming is like cooking from a book. Just as recipes give directions for measuring and mixing ingredients, programs give computers step-by-step instructions to follow. The new weather model built by Anandkumar and her team uses a kind of artificial intelligence called machine learning, which is more like cooking from scratch and developing a recipe. Machine learning lets computers solve problems through experience, much as people do.
Machine learning excels at combing through reams of data to find patterns, and is widely used in everyday life. Examples include detecting fraud from spam emails to credit card theft. Hedge funds also use machine learning to gauge the success of companies based on how many cars are in their parking lots.
Unlike conventional weather models, the new AI weather model has the capacity to handle historical as well as current weather data. “It learns from the past to predict the future,” Anandkumar explains. “There’s a wealth of hourly data since the 1970s that we’re barely making use of.”
The AI model rivals numerical models in accuracy, is cheaper because it doesn’t require a supercomputer, and is “lightning fast — it only takes a second,” Anandkumar says. Her team reported their findings in a study submitted for publication.
The National Oceanographic and Atmospheric Administration, which includes the National Weather Service, is currently evaluating the team’s AI weather model. “We’re collaborating with them to help modernize their system,” she says.
Now the team is extending their model to predict the impacts of climate change as well as near-term weather. While climate change models are global, the new AI model’s affordability and astonishing speed lend it to local assessments of future risks. “The regional level is what matters for flooding and high temperatures,” Anandkumar says. “How extreme will the extremes be?”
Now that this year’s barrage of storms has let up at last, all eyes are on the colossal mountain snowpack — one of the highest on record — they left behind. Will the state face another round of devastating floods as snow melts in a rush this spring? And can this gush of water refill the shockingly depleted aquifers in the Central Valley, where farmers have over pumped groundwater for a century? Jeff Dozier, a UC Santa Barbara emeritus snow hydrologist, just wrapped up a three-year project called “Headwaters to Groundwater: Resources in a Changing Climate” that will help answer these questions.
“One of the biggest effects of climate change in California is on the Sierra Nevada,” Dozier says. “That’s where most of the agricultural water supply comes from, and agriculture is the biggest water user.” Irrigating California’s $50 billion worth of crops accounts for about four-fifths of water use statewide.
This year’s impressive snowfall notwithstanding, the state Department of Water Resources projects that the Sierra Nevada snowpack will shrink up to 65% by 2100 as warming shifts the line between rain and snow. “We’re transitioning to more rain at higher elevations than normal,” Dozier says. “The problem is that there’s no place to put it.” The state’s reservoirs only have room to capture the runoff from an average water year. “You have to worry about flood control,” he continues.
Just about everybody’s favorite solution is to put this “extra” water where it can do good instead of harm. Efforts are underway to inundate fields and bank floodwaters in the Central Valley’s vast underground storage basins. Called managed aquifer recharge, this strategy would also help bring the basins back into balance by 2042 as required under the 2014 Sustainability Groundwater Management Act.
Managed aquifer recharge may sound straightforward but it’s not. One challenge is that the bulk of the snowmelt flows into the Sacramento Valley, which is in the north, but the aquifers in direst straits are in the San Joaquin Valley, which is in the south. In a 2020 study, Dozier and colleagues found that managed aquifer recharge could basically solve groundwater troubles in the Sacramento Valley. But it would barely make a dent in the San Joaquin Valley’s overdraft unless water is imported from the north via the Sacramento-San Joaquin River Delta, and even that would only be a partial fix.
The idea of transferring snowmelt from north to south brings up another barrier to managed aquifer recharge: uncertainty over the water rights. “The unpleasant surprise we’re coping with right now is that the governance of water management in California is not suited to expediting decisions,” Dozier says. “When we have a lot of extra surface water, the question is who owns it and where can you divert it?”
The state wants to own that water itself. The Department of Water Resources and the State Water Board plan to develop a more “equitable approach for allocation of water rights for groundwater recharge” with an initial focus on the state “securing all reasonably available future flood flows in the Central Valley,” according to a 2022 state report called California’s Water Supply Strategy.
Water as a Way of Life
L’Eaux Stewart, Chairperson of the Big Pine Paiute Tribe, is all for more equitable groundwater recharge. She lives in the Owens Valley, high desert just east of the Sierra Nevada. This arid land gets only a few inches of rainfall each year because the soaring peaks cast a rain shadow over it. But the mountains also provide snowmelt that once kept the valley’s water table high and streams flowing through the hot, dry summers.
“We’re a land of little rain but not of little water,” Stewart says.
But, as warming diminishes the snowpack and melts it earlier and faster, the Owens Valley is becoming a land of little water too. “The snowpack is erratic and our springs are drying out,” Stewart says.
Wet years like this one aren’t much help. “We never get to heal and recover our landscape,” Stewart says. “When the snowpack is high, the water goes to the Los Angeles Department of Water and Power.” The agency has pumped Owens Valley groundwater and piped it the 200 miles to Los Angeles since the early 1900s.
Climate change on top of groundwater export threatens Stewart’s way of life. “I grew up in an incredibly traditional household, surrounded by the language, stories and practices,” she says. “About half of our cultural activities depend on water — losing it is crippling us.”
Stewart learned to build houses out of tall rushes called tules, and to tend willow patches so they bore the long stems needed for weaving baskets. Now the disappearance of spring-fed wet meadows and marshes has made these water-loving plants scarce where she lives.
Lack of water also hampers preparation of traditional foods such as acorns, which Owens Valley tribes once gathered in great quantities in the fall to help sustain themselves through the winter. They then buried baskets full of acorns in creek beds, where the cold, flowing water leached out the toxic, bitter tannins while keeping the acorns fresh.
“Water is more than precious,” Stewart says. “Without it, how can we keep our traditional activities alive and carry our culture into the future?”
Just as water rights were set long ago and may be due for a rethink, California’s water supply system could use modernizing to get ready for a future with both super wet and super dry years. The state has long had the most varied weather conditions in the country. “It’s always been a roll of the dice from year to year; so the system is geared to variability,” says John Leahigh, a civil engineer and water manager with the Department of Water Resources’ State Water Project, which was built in the 1960s to manage flood risks and to deliver water from Northern California to the San Joaquin Valley and Southern California. “But with climate change we’re expecting to see even more extremes.”
These extremes have already begun. “There’s even more variation than we’d seen historically,” says Leahigh, who has worked in State Water Project operations and management for three decades. “There’s more of the whiplash effect.” This winter’s abrupt switch from bone dry to overflowing is a perfect example of whiplash weather.
Water system updates underway or in the works include a mix of new technology, new infrastructure, and new ways of managing the infrastructure in place. Examples include the recently launched Airborne Snow Observatory program, which uses aircraft equipped with lidar to measure snow depths from a height of 23,000 feet.
“The Airborne Snow Observatory gives us a much better snowpack forecast,” Leahigh says, explaining that the traditional method relied on surveying snow manually at fixed points while the lidar scans entire watersheds. Accurate snowpack estimates are essential to forecasting the snowmelt that provides much of California’s water.
As warming shrinks snowpacks, however, the state will rely more on rainfall. However, current infrastructure is designed to collect snowmelt, which comes gradually, rather than storm runoff, which comes all at once. To take advantage of the higher river flows from intensifying storms, the state proposes building a 45-mile tunnel under the Delta. Called the Delta Conveyance, the tunnel would divert water from the Sacramento River and deliver it to the California Aqueduct, which runs nearly 450 miles from the Delta to Los Angeles. “When water is available through direct runoff, we need a way to capture it,” Leahigh says.
Another key climate adaptation uses recent advances in atmospheric river forecasting to optimize reservoir management, an approach aptly called Forecast-Informed Reservoir Operations (FIRO). Currently, water is released from reservoirs to make room for stormwater according to a federally-mandated schedule. This is a set timetable that is based on historical weather patterns rather than actual conditions, so releases are made whether or not a storm is incoming.
In dry winters, this can mean needlessly losing water that never gets restored. In contrast, FIRO tells reservoir operators when to “fill or spill,” Leahigh says. “Rather than drawing down blindly, they can hold onto water.”
These are just some of the ways big and small that the state can reshape water management. Others include boosting local supplies by, for instance, recycling wastewater and storing more groundwater near cities. Realizing how much can be done to prepare gives me hope that, even as the dice become loaded toward wetter wet and drier dry periods, California can roll with the uncertainty.